The Nineteenth Byte

PhiNotPi

3:01 PM

Yeah, they tend to freeze up when enemies are in sight.

@Calvin'sHobbies I would like to petition the increase of the 15ms time limit.

Nathan Merrill

hey has anybody here read my next codebots challenge?

I need to discuss a problem I have

TheBestOne

@NathanMerrill I have. But you might have changed it since.

Nathan Merrill

No its fine

So, one of the key aspects that I've included is that if a bot attempts to read and write a variable at the same time, the reader's result will be a random value

however, this becomes tough with if statements

lets say an if statement tests for var A, and if it is equal to B, writes to A

because all actions are theoretically happening at the same time, the A variable should be messed up, but if that is true, then it shouldn't have written to A in the first place

so, there are several options I have

1. If statements now take a turn to execute

2. All If statements now execute at the beginning

(and therefore, can't be messed up)

TheBestOne

3:22 PM

I would do option 1

Nathan Merrill

Ok, then do I simply move the pointer to line number -1?

Geobits

@PhiNotPi Instead of tagging along with your formation, would it be more helpful if I just ran a searchbot, constantly broadcasting the last known location of all enemies?

TheBestOne

@NathanMerrill Actually, I think option 2 would be cooler, but harder to implement.

@NathanMerrill What for?

Nathan Merrill

3:40 PM

If I do option 1, then I essentially have to take the thread variable, and reassign it to N-1

because at the end of the turn, the thread variable will increment by 1

however, if I'm on thread D, and my if statement uses D in its check, then this is still a problem

but that problem still exists with 1 or 2

I could just remove If statements :P

fundamentally the problem is that If statements follow the pattern read-write

TheBestOne

@NathanMerrill Ah. I see. I thought you meant the line before the first line in a bot.

Nathan Merrill

no

while the sequence of actions should be write-read

(to detect for errors)

TheBestOne

You could make it so the thread is exterminated if it does that.

And both variables get random values.

Nathan Merrill

if it does what?

if it uses the thread variable in its check?

TheBestOne

32 mins ago, by Nathan Merrill

lets say an if statement tests for var A, and if it is equal to B, writes to A

Nathan Merrill

3:48 PM

ah

that's really hard to detect

Runer112

not this time, J

I will defend the honor of CJam!

TheBestOne

@NathanMerrill Just keep a boolean[] array, for each thread, for each turn, of variables its read from or written to; and a list of line numbers.

Nathan Merrill

it would have to be bot-specific

Geobits

I got a tumbleweed on android.se :(

Nathan Merrill

because if thread 1 writes to variable A, and thread 2 uses it in its if statement

or lets say they both have if statements, one reads A, then writes B, the other vice versa

Runer112

3:53 PM

BrainSteel's gunning for a tumbleweed on codegolf

I hope (sort of?) that he makes it

TheBestOne

Nevermind then.

Nathan Merrill

Ok. What if, at the beginning of each turn, I identify all lines that are going be executed. If any of them are "if" statements, then I do the checks

then, I move the thread variable to the appropriate spot

because if statements are only writing to their own thread variable, it should work

PhiNotPi

4:08 PM

@Sp3000 I've deon som debugging by increasing the time limit on the sandbox.

It appears that the pixels don't always store their new position in the message.

Actually, nevermind.

Still investigating.

Wait, yes, that is happening.

Calvin's Hobbies

@PhiNotPi Do you mean the time limit in meta or the real question?

PhiNotPi

The meta question.

Calvin's Hobbies

I think I forgot to remove the edit time limit last time I updated meta. Sorry about that.

I can make the meta 15ms time limit longer too,

PhiNotPi

That's what I'm talking about, I increased the 15ms in meta to something much higher.

Because Sp3000 and I were doing a ton of debugging, but the debugging tools themselves caused the bot to go over the 15ms limit.

@Calvin'sHobbies Ohh, I get what you are talking about now.

I do kinda want the timie limit in the real thing to be increased.

Because it seemed like our bots were timing out a lot.

But I think it is the debugging tools which caused that.

Calvin's Hobbies

Yeah, the debug messages make things run a lot slower.

PhiNotPi

4:22 PM

As of right now, I don't think there is a need to do anything.

If we do all the optimizations we can on on our bots, and can't get it under the time limit, that's when I would want a raised time limit. They're not really doing anything crazy.

Calvin's Hobbies

Alright, good. I made the meta move time limit 1 second if it helps.

Maria Tidal Tug

@Calvin'sHobbies what happened to row column diagonal block?

Calvin's Hobbies

I figured getting a nearly optimal score of N = 8 or 9 would be too easy. I'll undelete it after a few revisions.

@PhiNotPi Are you and @Sp3000 gonna answer soon? I'm starting to feel I should have not provided so many testing areas ;)

PhiNotPi

4:40 PM

Our bots are literally riddled with bugs.

Like, they won't even function right.

Which I why I've pent literally all day on getting all the bugs out of it.

Sorry about taking so long with all of this.

Calvin's Hobbies

4:58 PM

Don't hesitate to tell me if you find any controller issues.

Doorknob

5:11 PM

Fun fact: Sandbox is the most flagged user on Programming Puzzles & Code Golf Meta.

4

Geobits

That's because all he posts is these incredibly broad list-type questions.

PhiNotPi

What are the flags for? Are they auto-flags for having such a massive post?

orlp

5:27 PM

Pretty proud of this answer:

0

A: Is the word type able with key adjacent to eachother?

Pyth, 66 ?"Yes".Am>2sm^-.uk2Cm.Dx"qwertyuiopasdfghjkl*zxcvbnm"b9.5dC,ztz"No I was surprised to learn Pyth doesn't have a hypothenuse function, so this will likely be beat by a different language. I'll propose a hypothenuse function to Pyth, so this atrocity won't happen in the future. Explana...

Runer112

nice

the idea of actually computing key distance crossed my mind, but I thought that would be harder for some reason

you made it clear it's not :p

orlp

depends if your language has floating point divmod

Runer112

CJam does, apparently

orlp

does it have hypothenuse?

Runer112

nope

orlp

5:31 PM

hmm

then I stand a chance

if it had hypothenuse I would almost surely get beat by a translation

Runer112

can you just check if the manhattan distance is <2?

orlp

that was my previous attempt at 64 bytes

I quickly deleted it

Runer112

does that not work?

orlp

qs

Runer112

that looks like a manhattan distance of 2.5 to me..?

orlp

5:34 PM

oh, you mean still with 0.5 offsets each row

Runer112

yes

orlp

my original attempt was just shifted to 0 0 1 :)

hmm

Runer112

should save you the distance squaring

orlp

I guess that should work

Runer112

oh wait, that's no good anyways

orlp

5:34 PM

absolute

yeah

Runer112

because the squaring also serves to get rid of negatives

yes

unless you have a 1-char abs and 2-char square

CJam does, not sure about Pyth

orlp

well, abs is 2 characters

however, sorting is 1 character

so I can sort before subtracting

Runer112

that works

orlp

however, that means distances will be negative

so I'll have to end up adding another character to get -2

so no gain

however, there might still be a gain

I can compare each component seperately <= 1, no?

Runer112

yes

orlp

5:43 PM

I'm going to square the components

gets rid of the sign, and allows me to compare to an integer (2)

PhiNotPi

@Sp3000 I finally got the bots to work!!!!

It's glorious!

They actually move around and do stuff now!

orlp

@Runer112 It's not any shorter for me to compare components

oh well

Sieg

Mi prenis longa rigardi ĉe Esperanton.

Runer112

I'm fairly confident you'll win anyways :p

Sieg

Seems like fun.

Runer112

5:51 PM

unless anyone wants to be scummy and copy your approach in a langauge that does have a distance function

Sieg

@Runer112 N@R in Joe.

length of range :P

In fact that would be one off.

PhiNotPi

@Sp3000 I think the only thing left to do is improve the "filler code" into an actual search pattern.

orlp

I was thinking about cheating in the rat golf question

by only solving N = 1 and just drawing one path from a to A

it might just be the best scoring approach, considering the amount of logic needed to actually solve the problem

Runer112

what if I implement it correctly up to however many paths can fully fit in the space

and then say it works up to N=infinity

because any values larger than what can fit will print Impossible

orlp

because you can only implement up to N=26

the question does not allow higher values of N

Runer112

6:04 PM

heh, good point

this is probably a prime example for a question where you could make a brute force answer that won't finish before the heat death of the universe occurs

but you can probably prove is correct

orlp

I don't know

if you can do that in under 26*shortest_solution_n_1

the easiest would probably to write a simulator

and enumerate all possible boards

Runer112

yes, that's what I'm getting at

orlp

but I still don't know if you can fit that under 26*shortest_solution_n_1

for the shortest solution all you have to do is create a ring and 2 separators

Runer112

paths have to be one character wide

and you don't get to choose where the start and end are

oh

I see what you mean

orlp

#######a######################################################
#     #                                                      #
# ########################################################## #
# #                                                        # #
# #                                                        # #
# #                                                        # #
# #                                                        # #
# #                                                        # #

Runer112

6:13 PM

yeah

orlp

the ring is always static

the separators are reasonably easy

I was working on a solution that builds the ring and handles the separators really easily by rotating the board 4 times

Runer112

I just generally avoid questions like that, I couldn't keep track of the complexity :p

randomra

Haven't found any simple 6-neighbourhood checker expression for the word-typing question. :/

orlp

@randomra did you check out my answer?

randomra

@orlp Yes, I didn't want to copy that, but find a general is6neighbour(x1,y1,x2,y2) function

orlp

6:22 PM

@randomra is6neighbour is simply distance

randomra

I meant in a simple square grid

orlp

you can turn the simple square grid into the slanted grid rather easily

xn += yn/2

randomra

that falls into the long expression category (at least how I wrote it in J)

Optimizer

6:41 PM

I am only gone for a couple of hours and user23013 gets beaten in CJam ?

Runer112

oops?

we both got rekt by Pyth anyways, though

although that's probably largely due to a superior algorithm

Optimizer

thats a different issue.

Runer112

oh

Optimizer

@Runer112 pyth has so many advantages even if the algorithm is same

Runer112

yeah Pyth is the new OP

orlp

6:44 PM

well, I developed almost all of the new Pyth features :P

Optimizer

go away

Runer112

I'll continue to take solace in the belief that Pyth still can't compete in questions that require a more imperative-style answer

which may or may not actually be true

Optimizer

or a stack

orlp

imperative is always a bit painful

because with imperative you have to make the operands explicit

Runer112

that's kind of what I meant, the ability to not write functionally

orlp

6:45 PM

in stack based languages you don't

I would have a really good answer to the rat question if I could use my latest feature of Pyth

but I can't use features I developed after the question was posted :P

would be unfair

Runer112

solve the rat question built-in? :P

orlp

nah

not that bad

it's an indexed substitution function

so [2 3 5 3] [1 2] [8 8] becomes [2 8 8 3]

Pyth also isn't great for 2D questions

Runer112

I had an idea about how to make the most optimized golfing langauge possible

orlp

step 1: ready source in binary

Nathan Merrill

I have a KoTH idea, but I feel like its been done

1. all players pick a number from 1 to N

2. all duplicates receive 0 points

3. all other players receive the number that they picked

4. repeat

PhiNotPi

6:59 PM

There's codegolf.stackexchange.com/questions/4834/…

Which is slightly different, but has the same idea of picking a number and avoiding duplicates.

Nathan Merrill

right, I saw that

PhiNotPi

By the way, my answer there helped lead to some of the KOTH rules we have today.

Nathan Merrill

but that's slightly different in that only 1 person gets points

randomra

@NathanMerrill my concern is that if someone writes a print(N), if everyone is reasonable, he wins

I elaborated on this in a sandbox question I think

PhiNotPi

And then there's the Morra game which has the mechanic of choosing your own reward, but the rest of it is entirely different.

orlp

7:03 PM

@randomra but everyone knows htis

@randomra so everyone will pick N

Nathan Merrill

I would allow suicidal bots

orlp

then a metagame evolves where people pick N-1 to defeat the N'ers

randomra

@orlp if you saw some posted print N you will not do the same

Nathan Merrill

If I were to write a bot that took the highest value posted last, and print it, it should beat those types of bots

orlp

@randomra No, everyone would start writing programs that randomly choose another number, weighted

if you'd only pick N you'd lose

because the people who'd choose weighted numbers would own you

@randomra your argument holds in a 2 player game, but not above that

Nathan Merrill

7:06 PM

oh! I know!

you can't pick the same number twice in a game

and you would have N turns

orlp

not interesting

Nathan Merrill

N*2

that makes more sense

orlp

make the number of turns a small percentage of N

Nathan Merrill

bah, I was thinking backwards

you are right

N/2

orlp

no

n/100

Nathan Merrill

7:08 PM

eh, not a huge difference between N/2 and N/100

orlp

or n/10

something in between

Nathan Merrill

the same strategies will occur

randomra

or everyone can veto a number in every turn which can't score

Nathan Merrill

that's not a bad mechanic

you pick two numbers, one is your score, the other your veto

duplicate scores are allowed

orlp

either way, I think KOTH in SO format is kind of flawed

Nathan Merrill

7:10 PM

KoTHs on SO get lots of views

orlp

you shouldn't be able to see the source of competitors until post-mortem

Nathan Merrill

I'm totally ok with seeing the source

orlp

it depends on the challenge

in some it is fine

in a vast majority it leads to instant optimal counterstrategies, and forces you to post at the last second if you want to win

Nathan Merrill

the only difference is whether they are instant or not

by watching a KoTH enough, you'd be able to identify what they are doing and develop a counterstrategy

orlp

no, if you don't have access to the source you might never be able to fully figure out the behaviour of a competitor

Nathan Merrill

7:12 PM

sure, but you don't need to fully figure it out

just enough to counter it

and if your counter doesn't work, you'll figure out why in how it interacts with the competitor

KoTHs are all about building strategies that counter other ones and are tough to counter

IMO, it doesn't make sense to change them to make them hard to observe

New Meta Posts

7:44 PM

0

Q: How do I find the answer number in a post?

I've seen pages like this where they have links to answers in the description. I've noticed this works by adding a number to the end of the URL (i.e, 12345). How do I find this number so I can link things? (I know how to do it for my answers, but I'm wondering about answers in general.)

support

New Sandboxed Posts

8:08 PM

0

A: Sandbox for Proposed Challenges

Find sequences in RSA_1024 I found an interesting way to factor numbers, while its far from been any efficient in compare to other known methods, its lots of fun! Lets assume we want to factor N. X will be its factor, if and only if N/X = floor(N/X) So theoretically, if you could find a diff...

Pietu1998

Someone should make a KoTH where the objective is to die, with no suicide action. Pretty sure suicidal submissions wouldn't be banned :P

PhiNotPi

That might be interesting.

And players would have to be forced to harm each other.

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

Prepare to... die? code-golf game number Background An occasional source of mild vexation in tabletop role-playing games is dealing with rolls involving many dice. Casting a Disintegration spell may be instantaneous, but rolling and adding together 40 dice certainly isn't! A number of sugge...

Pietu1998

>If there is no solution (as r is too small), the program should output the string "I AM A SEXY SHOELESS GOD OF WAR!".

New Meta Posts

0

Q: Is it OK to link other people's answers?

I was on this page, and I noticed some entries for the list in the description were missing. Is it OK for me to link these entries, or do I have to ask permission from the posters?

discussion

Ilya_Gazman

8:19 PM

hello

So we have a sandbox room, or can I ask things about sandbox here?

PhiNotPi

Ask in here.

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

The Sound of Words People have written many things allowing one thing to be visualized as another. Now I propose being able to translate letters to music! Its your job to write a program that takes a text file and outputs a sound file with each letter converted to a specific note from C3-C8. Di...

Ilya_Gazman

@PhiNotPi I just added newquestion for sandbox, can you please add it to the list. I can't edit there.

0

A: Sandbox for Proposed Challenges

Find sequences in RSA_1024 I found an interesting way to factor numbers, while its far from been any efficient in compare to other known methods, its lots of fun! Lets assume we want to factor N. X will be its factor, if and only if N/X = floor(N/X) So theoretically, if you could find a diff...

PhiNotPi

@Ilya_Gazman Since it's not strict code-golf, I'm putting it under "other."

Ilya_Gazman

@PhiNotPi Yeah sure. Its not golf.

tnx

PhiNotPi

8:27 PM

done

Ilya_Gazman

@PhiNotPi tnx. Btw, do you like it?

PhiNotPi

I would suggest an example output that shows the first 5 to 10 lines of output.

There;s not really a good definition of what counts as a sequence.

Ilya_Gazman

well, that could be what ever

the goal is to have the biggest sequences possible.

May be the definition is that I can use your sequence to factor N. But I am afraid that it will only make the things to be more unclear.

PhiNotPi

8:46 PM

Also, you should probably pick a single winning criterion. The four you have listed are pretty much mutually conflicting:

Range been covered by sequences.
Performance
Shortest code
Submission time.

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

Code Golf - Roman Numeral Converter You're most likely familiar with Roman Numerals. So, the challenge is simple - build a program that converts Arabic numerals into Roman ones. Here is some sample input/output (the prompt is :, you don't need this): : 1 I : 20 XX : 42 XLII : 100 C For refere...

Ilya_Gazman

Edited

BrainSteel

9:14 PM

Eyo, folks. I've been thinking a lot about this challenge. I have the notion that almost all possible inputs are indeed solvable, but I can't find a general solution for proving it one way or another. Has anyone determined solvability accurately?

Optimizer

@Calvin'sHobbies can't your controller be tweaked a bit to accept a code block in order for new answers to not use sandbox post for testing their submissions

randomra

@NathanMerrill did you solved the problem codeforces.com/contest/6/problem/D ? or do you think my solution (chat.stackexchange.com/transcript/message/20885905#20885905) is correct?

xnor

I was going to make a code-golf about finding the multiple of a number with the smallest digit sum in polynomial time, but now I think it's related to a cryptographically hard problem

randomra

@Optimizer there is this: imgur.com/fyknZ5T

Optimizer

oh then why does the sandbox post still exist ? :)

randomra

9:26 PM

doesn't hurt, maybe CH can test there updates

Calvin's Hobbies

Just as a staging area. Doorknob said it could remain open.

Optimizer

do i get a gold badge every 100 consecutive days ?

Calvin's Hobbies

Doubt it. It doesn't say "This badge can be awarded multiple times."

Optimizer

what a waste

Calvin's Hobbies

And you'd have like 5 Enthusiasts

randomra

9:54 PM

if ones name is not offered with @, I can't ping him, right?

randomra - correct

not correct

it doesn't ping

the auto-correction is client-side, and pinging is server-side. they don't always agree. after the last update to auto-correct it's much more reliable but not 100% correct.

at the very least, you should try hard-reloading the chat (that sometimes adds a few names to the auto-correct list)

however, if it's really important, you can ask a mod to superping them. in that case, they'll get a notification even if they've never used chat before.

Optimizer

wasn't his question simply that not putting an @ before a user name will not ping him/her ?

Martin Büttner

10:01 PM

I interpreted "name is not offered with @" as "name doesn't show up in auto-complete suggestions". but who knows.

randomra

@MartinBüttner that's what I meant

Martin Büttner

@Optimizer no.

randomra

e.g. I can't ping @FUZxxl

Calvin's Hobbies

Martin, I reckon the asyc koth should be unstickied since it's not going anywhere and the fortnightly challenges are on hiatus

Martin Büttner

sure

Optimizer

10:02 PM

@Calvin'sHobbies column vs col

Martin Büttner

@randomra looks like he hasn't been in chat for a month. I think the limit is 2 weeks, but I have no clue why I think that, so it could be completely made up.

Calvin's Hobbies

Thanks

randomra

@MartinBüttner Thanks, I see

Calvin's Hobbies

Would codegolf.stackexchange.com/questions/48401/… be better if i made distinct chars then lowest N be the win criterion?

Answers would be more interesting

@Optimizer Sorry, changed it

Optimizer

space is included in the count, right ?

Calvin's Hobbies

10:11 PM